Associations of baseline, cumulative, and dynamic trajectories of the cardiometabolic-frailty index with incident cardiovascular disease: a longitudinal analysis based on the China Health and Retirement Longitudinal Study (CHARLS)

Background Cardiovascular disease (CVD) remains the leading cause of death worldwide [1]. Both cardiometabolic dysregulation and frailty are independently associated with increased CVD risk [2,3], yet an integrated marker capturing both dimensions has not been fully evaluated. We constructed the cardiometabolic-frailty index (CMIFI) as the product of the cardiometabolic index (CMI) and the frailty index (FI), and examined its baseline levels, cumulative exposure, and longitudinal trajectories in relation to incident CVD. Methods We used data from CHARLS (2011/2012–2020) [4]. Adults aged > = 45 years without baseline CVD and with complete CMIFI and covariate data were included (n = 7,748). CMIFI was calculated as CMI x FI. Cumulative CMIFI was defined as mean(CMIFI_2011, CMIFI_2015) x 3 years. Group-based trajectory modeling (GBTM) was the primary approach for identifying longitudinal CMIFI trajectories, with K-means clustering as an exploratory sensitivity check. Cox proportional hazards models, restricted cubic splines (RCS) [5], subgroup analyses, and CMI x FI interaction were performed. An illustrative prediction analysis examined discrimination, calibration, decision curve analysis (DCA), and SHAP values. Sensitivity analyses included standardized CMIFI, separate heart disease and stroke outcomes, stratified Cox regression, inverse probability weighting (IPW) Cox [7], exclusion of baseline diseases and early events, reclassification of smoking/drinking, competing-risk scenarios under assumed death hazards, and over-adjustment models excluding lipid variables. Cardiometabolic multimorbidity (CMM) was evaluated as a secondary outcome. Results Among 7,748 participants, 1,750 incident CVD events occurred during a median follow-up of 9 years. Compared with the lowest quartile (Q1), the highest baseline CMIFI quartile (Q4) was associated with a 1.72-fold increased risk of CVD (95% CI 1.47–2.01; P for trend < 0.001) after full adjustment. Each 1-SD higher baseline CMIFI was associated with a 1.11-fold higher CVD risk (95% CI 1.07–1.15). GBTM identified two trajectories in the longitudinal subsample (n = 3,495): low-stable (n = 3,420) and high-declining (n = 75). Compared with the low-stable group, the high-declining group had a 1.84-fold (95% CI 1.28–2.65) higher CVD risk after full adjustment; the bootstrap 95% CI was 1.29–2.61. Cumulative CMIFI showed a stronger dose-response: Q4 versus Q1 HR 2.99 (95% CI 2.42–3.69). CMI and FI showed a sub-multiplicative joint effect (multiplicative interaction HR 0.79, 95% CI 0.64–0.97; RERI − 0.29, 95% CI -0.56 to -0.01), consistent with the product form of CMIFI as a pragmatic simplification. An illustrative prediction model achieved an AUC of 0.671 in the validation set; NRI and IDI were null to small and are reported in Supplementary Table S7. DCA is shown in Fig. 6 and SHAP in Supplementary Figure S5. As a secondary outcome, incident CMM occurred in 464 of 7,326 participants over 4 years; Q4 versus Q1 HR was 5.86 (95% CI 4.23–8.11). Results were robust across sensitivity analyses. Conclusions Higher baseline, cumulative, and high-declining CMIFI trajectories were independently associated with increased CVD risk in Chinese middle-aged and older adults. CMIFI was also associated with incident cardiometabolic multimorbidity. CMIFI may serve as a simple, integrative indicator for CVD risk stratification and long-term monitoring.